Structural basis for regulation of the nucleo-cytoplasmic distribution of Bag6 by TRC35

Jee-Young Mocka, Yue Xub, Yihong Yeb, and William M. Clemons Jr.a,1

aDivision of Chemistry and Chemical Engineering, California Institute of Technology, Pasadena, CA 91125; and bLaboratory of Molecular Biology, National Institute of Diabetes and Digestive and Kidney Diseases, National Institute of Health, Bethesda, MD 75534

Edited by Jeffrey L. Brodsky, University of Pittsburgh, Pittsburgh, PA, and accepted by Editorial Board Member F. Ulrich Hartl, September 20, 2017 (received for review February 23, 2017) The metazoan BCL2-associated athanogene cochaperone 6 heterotetrameric Get4–Get5 complex (3, 13), but the addition of (Bag6) forms a hetero-trimeric complex with ubiquitin-like 4A and Bag6 leads to different interactions. transmembrane domain recognition complex 35 (TRC35). This Crystal structures of fungal Get4 revealed 14 α-helices that can Bag6 complex is involved in tail-anchored protein targeting and be divided into N-terminal domains (NTD) and C-terminal do- various protein quality-control pathways in the cytosol as well as mains (CTD) (Fig. 1A and Fig. S1A)(14–16). The NTD contains regulating transcription and histone methylation in the nucleus. conserved residues (Fig. S1C) involved in binding and regulating Here we present a crystal structure of Bag6 and its cytoplasmic Get3 (13, 15), while the CTD forms a stable interaction with the retention factor TRC35, revealing that TRC35 is remarkably Get5 N terminus (Get5-N) (15). The Get5 interface includes a conserved throughout the opisthokont lineage except at the hydrophobic cleft in Get4 formed by a β-hairpin between α13 and C-terminal Bag6-binding groove, which evolved to accommodate α14 that binds the helix in Get5-N (15). Sequence alignment of Bag6, a unique metazoan factor. While TRC35 and its fungal ho- Get4/TRC35 and Get5/Ubl4A led to predictions that the Get4 molog, guided entry of tail-anchored protein 4 (Get4), utilize a β-hairpin and Get5 N-terminal extension are absent in metazoans conserved hydrophobic patch to bind their respective partners, (Fig. 1 A and B) (15), which suggested a different structural or- Bag6 wraps around TRC35 on the opposite face relative to the ganization for the metazoan Bag6–TRC35–Ubl4A complex. Get4–5 interface. We further demonstrate that TRC35 binding is In metazoans, Bag6 abrogates the interaction of TRC35 with critical not only for occluding the Bag6 nuclear localization se- Ubl4A by separately binding these factors (3, 17) resulting in a BIOCHEMISTRY quence from karyopherin α to retain Bag6 in the cytosol but also heterotrimer. The interface between TRC35 and Bag6 had pre- for preventing TRC35 from succumbing to RNF126-mediated ubiq- viously been mapped to the CTD of TRC35 and the region on uitylation and degradation. The results provide a mechanism for Bag6 that contains the bipartite nuclear localization sequence regulation of Bag6 nuclear localization and the functional integrity (NLS) (3, 18). Overexpression of TRC35 results in Bag6 retention of the Bag6 complex in the cytosol. in the cytosol (18), suggesting that TRC35 plays a role in de- termining the nucleo-cytoplasmic distribution of Bag6. Potential tail-anchor targeting | X-ray crystallography | tail-anchor recognition roles for Bag6 in the nucleus are varied and include modulation of complex | proteasome-dependent degradation | GET pathway p300 acetylation (19, 20), facilitation of histone methylation (21), and regulation of DNA damage signaling-mediated cell death ail-anchored (TA) membrane contain a single, (22). How Bag6 localization is regulated remains unclear. TC-terminal, hydrophobic transmembrane-helix domain (TMD) In this study, we present a crystal structure of a complex that (1). The TMD, which acts as the endoplasmic reticulum (ER) includes TRC35 from Homo sapiens (Hs) and the TRC35- targeting signal, emerges from the ribosome after translation binding domain of HsBag6. The structure of TRC35 reveals termination necessitating posttranslational targeting and in- sertion. Transmembrane domain recognition complex (TRC) Significance proteins in mammals and guided entry of tail-anchored proteins (GET) in fungi are the best-characterized molecular players in The metazoan protein BCL-2–associated athanogene cochaper- TA targeting (2). In the TRC pathway, TA proteins are captured one 6 (Bag6) acts as a central hub for several essential cellular by the small glutamine-rich tetratricopeptide-repeat protein A processes, including immunoregulation, regulation, apo- (SGTA) after release from the ribosome then transferred to the ptosis, and proteostasis. These roles are in both the nucleus and ATPase TRC40, a process facilitated by the hetero-trimeric Bag6– the cytosol, but the mechanism by which Bag6 traffics between TRC35–Ubl4A complex (3, 4). TA binding stimulates hydrolysis of these compartments remains elusive. Here we present the crystal ATP by TRC40 (5) leading to release of the TRC40–TA complex structure of Bag6 in complex with its cytoplasmic retention factor that then localizes to the ER membrane via its interaction with transmembrane domain recognition complex 35 (TRC35) and suggest a mechanism of regulation for the nucleo-cytoplasmic calcium-modulating ligand (CAML) (6). Release of the TA sub- transport of Bag6. strate for insertion is facilitated by CAML and tryptophan rich basic

protein (WRB) (7). Notably, loss of WRB in mouse cardiomyocytes Author contributions: J.-Y.M., Y.Y., and W.M.C. designed research; J.-Y.M., Y.X., and Y.Y. and hepatocytes results in partial mislocalization of TA proteins (8), performed research; J.-Y.M., Y.X., and Y.Y. contributed new reagents/analytic tools; suggesting that other pathways, including the signal-recognition J.-Y.M., Y.Y., and W.M.C. analyzed data; and J.-Y.M. and W.M.C. wrote the paper. particle (SRP) (9), SRP-independent targeting (SND) proteins The authors declare no conflict of interest. (10), the Hsc70 family of chaperones (11), and ubiquilins (12), can This article is a PNAS Direct Submission. J.L.B. is a guest editor invited by the Editorial compensate to target TA in vivo. Board. Structural information for the mammalian proteins is sparse; Published under the PNAS license. however, homology to the equivalent fungal counterparts, Get1– Data deposition: The atomic coordinates and structure factors reported in this paper have been deposited in the (PDB), https://www.wwpdb.org (PDB ID code 5 and Sgt2, allow for parallels. Bag6 is unique among the 6AU8). mammalian proteins, and its introduction alters the molecular 1To whom correspondence should be addressed. Email: [email protected]. architecture of the central hand-off complex. For TA targeting, This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. the heterotrimeric Bag6 complex is equivalent to the fungal 1073/pnas.1702940114/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1702940114 PNAS Early Edition | 1of6 Downloaded by guest on September 30, 2021 A α11 α12 Get5-N Get4 Hsap KSSASVVFTTYTQKHPSIEDGPPFVE------PLLNFIWFLLLAVDG 249 Aque PDEFGQLLVDLSYKEAKPHEIEMFIT------QAVLQLLVLEKKD 229 Spom LISAWNSLETFTKHFTKSNAPDVENMSFDGK----DFPVFKEYPQMNFLHLLIFTAYR 239 Ncas IRFVNQAKEAFLTSFINKYAPKVEITEKNLGSGVFKMYYFESYKSLNFLQLLVLTCQT 260 Scer ISFAHESKDIFLERFIEKFHPKYEKIDKNG----YEIVFFEDYSDLNFLQLLLITCQT 257 Afum LRNANKAFLVFTSRLSSSNTSLGVQEVSSAS---SDVRVFPSLPLLNFISMLLLTIQR 278 Fig. 1. The structure of the Bag6NLS–TRC35 hetero- Ncra VRAANTSYRAFVSALTAENSSLGVQNVESSQG--SDIRIFPSLPLLNLLGLLLLAVQK 256 Smus TRAANKALLLFTSRLSTSHPGLGVQSISSPS---SDIRIYPSLPLLNFLGLLLLAVQR 246 dimer. (A) The structure of Saccharomyces cerevisiae – B C N Get4 Get5N complex [Protein Data Bank (PDB) ID code: 16 3LKU], Get4 in rainbow and Get5 in magenta. Sequence alignment of TRC35/Get4 from Homo sapiens, Amphi- 15 N 6 medon queenslandica, Schizosaccharomyces pombe, C 13 14 4 2 N 8 Naumovozyma castelli, Saccharomyces cerevisiae, As- pergillus fumigatus, Neurospora crassa,andSphaerulina musiva. The secondary structure based on fungal Get4s 90° 12 10 5 3 1 9 7 is highlighted above the text. (B, Left) The overall structure of Bag6–TRC35 heterodimer in cylinder rep- C Bag6 TRC35 11 resentation with Bag6 in light pink and TRC35 in rain- NLS bow. The nuclear localization sequence is highlighted in α12 90° V254 ‘ ’ CDα13 α13 V1008 sticks on Bag6. (B, Right) A 90° in plane rotated bottom α12 M271 view. The 16 helices of TRC35 are labeled from N to C W1012 V257 M1040 terminus. (C) Comparison of the C-terminal faces of α11 F188 Y1036 TRC35 and Get4 that bind Bag6 and Get5, respectively. W1004 F242 L258 The arrows highlight the significant structural differ- I1016 L1032 ence in the residues between α11 and α12. (D) Zoomed α11 Y262 F195 in view of the regions, defined as interface I and II. – Interface I Interface II Hydrophobic residues that are involved in Bag6 Get5Get4 Bag6 TRC35 TRC35 dimerization are shown as sticks.

the conserved architecture relative to fungal Get4 homologs. are well conserved (Fig. S1 C and D). As predicted from se- Surprisingly, Bag6 utilizes the same conserved pockets on TRC35 quence alignment, the β-hairpin in Get4 is replaced by a shorter that are recognized by Get5 on Get4; yet the extended domain loop in TRC35 (Fig. 1C, arrows). In fact, sequence alignment of wraps around the opposite face of the α-solenoid. The resulting the predicted Get4 proteins of selected opisthokonts revealed interaction leads to masking of the bipartite NLS preventing nu- that the β-hairpin is unique to the fungal lineage (Fig. S2 and clear trafficking of Bag6 by blocking its binding to karyopherin α Table S3). The C-terminal α16 is flanked by two extended (KPNA), which is demonstrated experimentally. The interaction stretches that cover part of Bag6 (Fig. 1B). also prevents RNF126-mediated degradation of TRC35. These The Bag6 NLS region wraps around the TRC35 CTD (Fig. 1B, results suggest a mechanism for regulation of Bag6 nucleo- light pink) at an interface that resembles that of Get5–Get4 (Fig. cytoplasmic distribution. 1C). The interaction is stabilized by two hydrophobic interface sites. In interface I, Bag6 α1 and α2 dock into a conserved pocket The Crystal Structure of the TRC35–Bag6–NLS Complex created by α12 and α13 of TRC35 (Fig. S1D) and include W1004, A complex between TRC35 (residues 23–305), shown to be V1008 and W1012 from Bag6 and V254, V257, F242, L258 and competent for TA transfer (3), and a minimal TRC35-binding Y262 from TRC35 (Fig. 1D). In fungal Get4–5, the Get5 domain of Bag6 that contains the NLS (residues 1,000–1,054) N-terminal helix docks into a groove formed by α12, α13, and the was coexpressed, purified and crystallized. A dataset from a β-hairpin of Get4 (Fig. 1C). The missing β-hairpin in single crystal was collected to 1.8 Å resolution and phases were TRC35 results in the Bag6 helix shifting away from α11 toward obtained using single wavelength anomalous dispersion from a α12 and α13 near the bottom of TRC35 (Fig. 1C, arrows). The rubidium derivative. The final model refined to 1.8–39 Å had differences in the interface result in rearrangement of α11, α13, good statistics with an Rfree = 20.1%. Full crystallographic sta- and α15 of TRC35 (Fig. 1C) relative to those of Get4. Interface tistics are provided in Table S1. All TRC35 residues in the II is less extensive involving fewer residues, L1032, Y1036, crystallized construct were visible in the electron density except M1040 of Bag6 and F195, M271 of TRC35 (Fig. 1D). While the for S137 and G138 that are disordered in the loop between connecting loops between the two interfaces are well-ordered in α6 and α7. For Bag6, residues 1,002–1,043 were resolved with both contexts, the Bag6 loop makes fewer contacts to TRC35, only a few terminal residues ambiguous in the electron density. buried surface area of 2,841 Å2, than the extensive interactions in TRC35 has the same overall architecture as fungal Get4 (Fig. the Get5-loop to Get4, buried surface area of 5,554 Å2 (Fig. 1B). S1A), revealing that the Getfourfold has been conserved across The structure reveals how TRC35 masks the Bag6 NLS (Fig. Opisthokonta. Opisthokonta includes metazoan, choanozoan, 1B). The conserved first basic cluster of the predicted NLS and fungal lineages (Fig. S2), which share common ancestry as (K1024RVK) is sequestered between interfaces I and II (Fig. 1B, determined by analysis of 16S-like rRNA (23) and protein se- sticks). Only the first lysine residue of the second cluster α α quences (24). The first 15 -helices form an -solenoid fold that (K1043RRK) is resolved in our model (Fig. 1B, sticks). The can be divided into N- and C-terminal halves (NTD and CTD) truncated C-terminal residues of TRC35 in our construct are (Fig. 1B). Alignment of the NTD between TRC35 and Get4 predicted to be disordered and include conserved multiple neg- results in an RMSD = 1.380 Å, while the equivalent alignment in ative charges (five Glu and three Asp) (Fig. S3) that would be in = B the CTD results in an RMSD 3.429 Å (Fig. S1 ). As seen in position to mask the second NLS basic cluster, K1043RRK, Get4 (15), there is likely some flexibility between the two halves through charge-mediated interactions. based on differences across crystal forms. TRC35 and TRC40 are conserved throughout opisthokont evolution and seem to Probing the Bag6–TRC35 Interface occur as a pair, suggesting the essentiality of the two proteins in To validate the interfaces observed in the structure, we gener- the pathway (Fig. S2). Accordingly, residues at the TRC35– ated alanine mutants for analysis by yeast two-hybrid (Y2H) TRC40 interface (13) and the TRC35–Bag6 interface in TRC35 analysis. A fragment of Bag6 (residues 951–1,126) was attached

2of6 | www.pnas.org/cgi/doi/10.1073/pnas.1702940114 Mock et al. Downloaded by guest on September 30, 2021 to the GAL4 transcription activating domain and full-length bind Bag6 at its physiological binding site are unstable and become a TRC35 was attached to the GAL4 DNA-binding domain. The target for Bag6-mediated degradation. In this case, unassembled transformation of both vectors was confirmed by cell growth on TRC35 would bind to the Bag6 substrate-binding site. Because the SC-Ura-Leu medium (Fig. S4 A and B). Of the single Bag6 construct used in our Y2H studies lacks the substrate-binding substitutions (Fig. 2A and Fig. S4 A and B), only one residue domain, increased binding is not observed in the double mutant ex- TRC35 (Y262A) at the end of interface I (Fig. 1D) disrupted periments. If this model is correct, the higher molecular weight bands the Y2H interaction (Fig. S4A). The Bag6 mutations W1004A should be ubiquitylated TRC35 bound to the substrate-binding site and W1012A localize at interface I and Y1036A localizes at in- on Bag6. terface II. While the individual mutations did not show a pheno- To verify a role for Bag6 in TRC35 ubiquitylation, Bag6 de- type, the combination of the two mutations (W1004A/Y1036A ficient cells were used to coexpress TRC35· holdase FLAG, or W1012A/Y1036A) synthetically disrupted the interaction (Fig. Bag6 holdase GFP variants and HA holdase ubiquitin (Ub). 2A). For mutants that failed to grow, expression was confirmed by Proteins bound to Bag6 were first immunoprecipitated with GFP immunoblotting (Fig. S4C). The results confirm that both inter- Ab (Fig. S5 A and B,lanes2–7). To remove other ubiquitylated faces are critical for forming a stable complex between Bag6 Bag6 substrates (18, 27), the samples obtained from the GFP IP and TRC35. were subject to a second round of IP using anti-FLAG beads We further validated the interaction of TRC35 with full length under denaturing conditions (Fig. S5 A and B,lanes9–14). Im- Bag6 in the context of a mammalian cellular environment. GFP- munoblotting showed that all Bag6 variants pulled down ubiq- tagged WT Bag6 or the mutants Bag6(W1004A), Bag6(W1012), uitylated proteins, suggesting that the mutations did not affect its Bag6(Y1036A), Bag6(W1004A/Y1036A), and Bag6(W1012A/ substrate-binding activity (Fig. S5 A and B, lanes 2–7). After the Y1036A)werecoexpressedwithFLAG-taggedTRC35inCRISPR- second IP with FLAG Ab, TRC35 associated with wtBag6 carried a mediated Bag6-knockout 293T cells (25). Bag6 and associated small amount of Ub conjugates (Fig. S5 A and B, lane 9), while proteins were captured from cell extracts by immunoprecipitation those associated with Bag6 variants that disrupted physiological (IP) using a GFP-Ab. Viewed by immunoblot (Fig. 2B and Fig. S4 association with TRC35 carried significantly more Ub conju- D and E), each Bag6 mutation resulted in a reduction of the cap- gates (Fig. S5 A and B, lanes 10–14). Compared with single tured TRC35 relative to wtBag6 (lane 9 vs. lanes 10–12). The Bag6 mutations, TRC35 bound to the double mutants had the double mutants (lanes 13 and 14) also had reduced capture of highest Ub-modified to unmodified TRC35 ratio (Fig. S5 A and B, TRC35, but they unexpectedly appeared to capture more lanes 10–12 vs. lanes 13 and 14). Furthermore, the level of TRC35 than the single mutants. Furthermore, higher molecular TRC35 in cells expressing Bag6 double mutants was consistently BIOCHEMISTRY weight TRC35 bands appeared that became more pronounced lower than in cells expressing wtBag6 (Fig. S5C), which could be in the double mutants (Fig. 2, asterisk and Fig. S4 D and E). rescued by treatment with the proteasome inhibitor MG132 (Fig. This will be discussed below. Overall, the general effects of 3A, lane 2 vs. lane 4). Treating cells with MG132 also increased mutations at the Bag6 interface are consistent between the ubiquitylated TRC35 that was captured by either wtBag6 or a Y2H and the mammalian pulldowns. Bag6 double mutant (Fig. 3A). The accumulation of unmodified TRC35 (Fig. 3A, lanes 9 and 10 vs. lanes 11 and 12) is probably due Improperly Assembled TRC35 Is Ubiquitylated by RNF126 to depletion of free Ub in the cell (28). Collectively, these data As noted, IP results (Fig. 2B and Fig. S4 D and E) revealed two support the idea that the TRC35 molecules unable to associate with interesting observations: (i) an increase in TRC35 binding to Bag6 at the NLS domain become targets for Ub-dependent deg- Bag6 double mutants relative to single mutants (Fig. 2B and Fig. S4 radation through a client–chaperone interaction with Bag6. This D and E, lanes 11 and 12 vs. lanes 13 and 14) and (ii) the appearance suggests that the ubiquitylated TRC35 associates with the quality of higher molecular weight products of TRC35 captured by Bag6 control module (QC) of Bag6 (27). double mutants (Fig. 2B and Fig. S4 D and E,asterisk).Evidence To further ensure that unassembled TRC35 is a Bag6-QC supports that the stability of TRC35 requires forming a proper substrate, we investigated the effect of the Ub-ligase RNF126 on complex with Bag6 (17, 22). Given the implication of Bag6 as a TRC35 ubiquitylation. RNF126 is a Bag6-associated Ub-ligase chaperone holdase in protein quality control processes such as mis- utilized by the Bag6-QC for ubiquitylation of Bag6-associated localized protein degradation (26) and ER-associated protein deg- clients in the cytosol (27). If TRC35 ubiquitylation is mediated by radation pathway (18), we postulated that TRC35 mutants that fail to Bag6-QC, knocking down RNF126 in cells expressing Bag6 mutants would result in reduced ubiquitylation and increased stability of TRC35. To test this hypothesis, we examined the effect of siRNA- mediated RNF126 knockdown (KD) on TRC35 ubiquitylation. A B 5% INPUT IP GFP Bag6 (951-1126) 293T cells coexpressing HA·Ub, TRC35·FLAG, and either WT or W1004A/ mutant Bag6·GFP (W1004A/Y1036A or W1012A/Y1036A) were wt W1004A Y1036A Y1036A treated with siRNA against RNF126 followed by 6-h treatment with MG132. Bag6-associated TRC35 was isolated by two rounds of IP. Y1036A W1004A W1012A Vector wild type W1004A W1004A/Y1036A W1012A/Y1036A Vector wild type W1012A Y1036A W1012A/Y1036A W1012A/ W1004A/Y1036A Immunoblotting was used to analyze the relative ratio between 1234567 891011 12 13 14 Lanes TRC35 wt W1012A Y1036A Y1036A ubiquitylated TRC35 and unmodified TRC35 (Ub–TRC35/TRC35). 39 57 61 59 37 25IB GFP The ratio in cells expressing both wtTRC35 and wtBag6 without * RNF126 was defined as 1. In cells expressing Bag6 mutants, ubiq- IB FLAG uitylation of TRC35 bound to Bag6 is moderately increased (Fig. 3B, 3014015 14 20 22 lane 9 vs. lane 10); RNF126 KD reduces TRC35 ubiquitylation in Fig. 2. Validation of the Bag6–TRC35 interface. (A) Yeast two-hybrid assay cells expressing either wtBag6 or Bag6 double mutants, resulting in to validate the interface identified in the crystal structure. WT TRC35 con- approximately two- to threefold reduction in Ub–TRC35/TRC35 jugated to the DNA binding domain was expressed with WT or mutant Bag6 (Fig. 3B, lanes 9 and 10 vs. lanes 11 and 12). Importantly, like – (951 1126) conjugated to the transcription activating domain. Interaction MG132 treatment, RNF126 KD rescued the expression of TRC35 in was determined by ability to grow on adenine-negative medium. (B)WTor − − mutant Bag6•GFP was coexpressed in Bag6 / 293T cells with TRC35•FLAG cells expressing the double mutant (Fig. S6), presumably due to in- and immunoprecipitated using anti-GFP Ab. Amount of TRC35 retrieved by creased stability. Together these results demonstrate that ubiq- Bag6 was assessed by blotting with anti-FLAG Ab. The position of the higher uitylation of unassembled TRC35 is modulated by the QC role of molecular weight TRC35·FLAG is indicated by an asterisk. Bag6 and RNF126.

Mock et al. PNAS Early Edition | 3of6 Downloaded by guest on September 30, 2021 A 5% Input 1st IP: GFP 2nd IP: FLAG test this hypothesis, WT or mutant Bag6•GFP were expressed in -++ -+- MG132 Cos7 cells with or without TRC35•FLAG. The localization of wtBag6 Bag6 and TRC35 was examined by immunofluorescence. W1004A/Y1036A As expected (18), overexpressed wtBag6 and various Bag6 1 234 56789101112Lanes mutants were localized to the nucleus (Fig. 4A). This is likely due to excess Bag6 that cannot be retained in the cytosol by endog- enous TRC35. When TRC35 is coexpressed with wtBag6, the increased cytosolic pool of TRC35 captures wtBag6 and both A IB HA * IgG HC proteins stain primarily in the cytosol (Fig. 4 ). In accordance with the IP results (Fig. 2B), introduction of a single mutation— W1004A, W1012A, or Y1036A—results in some Bag6 localization * IgG LC to the nucleus (Fig. 4 B–D and Table S3). Mutations at both in- terface I (W1004A or W1012A) and interface II (Y1036A) further 12 17 35 36 14 19 42 41 3 7 9 13 disrupt the binding between Bag6 and TRC35 (Fig. 2), and Bag6 Bag6 localizes primarily to the nucleus (Fig. 4 E and F and Table S3). These results unequivocally establish TRC35 as a cytoplasmic TRC35 retention factor of Bag6. 10 1 28 15 50 3 44 19 36 1 59 25 ATF4 TRC35 Has Higher Affinity for Bag6 than Karyopherin-α 2 p97 All molecules destined for the nucleus must move through the nuclear pore complex, which spans the nuclear envelope and 5% Input 1st IP: GFP 2nd IP: FLAG B facilitates nucleo-cytoplasmic traffic (29). Macromolecules larger +-+-+- RNF126 siRNA than ∼40 kDa cannot freely diffuse through and require carrier wtBag6 α Bag6(W1004A/Y1036A) proteins, such as the karyopherin- (KPNA) family of trans- 1234 56789101112Lanes porters (30). The two basic clusters, R1024KVK and K1043RRK (Fig. S7A), in Bag6 are thought to act as a bipartite NLS by specifically recognizing the acidic substrate-binding surface of karyopherins (31). In HeLa cells, the K1045RtoS1045L mutation has been shown to abrogate nuclear localization of Bag6 (31). IB HA Therefore, TRC35 and KPNA likely bind Bag6 at the fragment that contains the NLS.

No TRC35 with TRC35-FLAG 34 44 24 23 57 49 26 17 13 18 5 6 Bag6 DAPI Bag6/DAPI Bag6(% Nuclear) TRC35-FLAG Bag6/TRC35/DAPI Bag6 A TRC35 WT

5 3 7 5 9 4168146136 6% RNF126 130.41Ub-TRC35 / TRC35 B (relative)

p97 W1004A 37% Fig. 3. Ubiquitylation of TRC35 in cells expressing Bag6 mutants is de- C − − pendent on RNF126. (A) Bag6 / 293T cells cotransfected with plasmids

encoding TRC35•FLAG (WT), Bag6·GFP (WT or mutant), and HA•Ub and W1012A treated with 10 μM MG132. The cell extracts were subject to two rounds of 65% IP with anti-GFP Ab then with anti-FLAG Ab in denaturing conditions. D − − Ubiquitylation was assessed by immunoblotting with HA Ab. (B) Bag6 / • 293T cells cotransfected with plasmids encoding TRC35 FLAG (WT), Y1036A Bag6•GFP (WT or mutants), and HA·Ub. The cells were simultaneously 64% treated with 10 μM MG132 and siRNAs against RNF126. The cell extracts E were subject to two rounds of denaturing IP with anti-GFP Ab and then with anti-FLAG Ab. TRC35 ubiquitylation was assessed by immunoblotting with W1004A /Y1036A HA Ab. 86% F /Y1036A TRC35 Masks the Bipartite Nuclear Localization Sequence of W1012A Bag6 and Retains Bag6 in the Cytosol 90% 10 m To investigate the Bag6–TRC35 interface in the context of Bag6 localization, the mutants identified from Y2H and IP were Fig. 4. Bag6 mutations at the TRC35 binding site results in nuclear locali- used for localization studies. Overexpression of TRC35 has been zation of Bag6. (A–F) Representative images of Cos7 cells expressing WT · · shown to retain Bag6 in the cytosol (18), suggesting that TRC35 FLAG and WT or mutant Bag6 GFP. Cos7 cells were transfected either with Bag6·GFP (WT or mutant) expressing plasmid alone or cotransfected TRC35 binding is required for cytosolic localization of Bag6. with TRC35•FLAG (WT) expressing plasmid. Cells were stained with anti-GFP Because the mutations identified in Y2H experiments specifically (green) and/or anti-FLAG (red) Abs. DNA was stained with DAPI where in- prevent TRC35 binding at its native binding site, exogenously dicated (blue). In the Bag6 column, the percent of cells that contain nuclear expressed Bag6 mutants defective in TRC35 binding would lo- Bag6 is noted in parenthesis. For each Bag6 variant, ∼100 cells were counted calize primarily in the nucleus regardless of TRC35 expression. To and categorized as having nuclear or cytosolic Bag6.

4of6 | www.pnas.org/cgi/doi/10.1073/pnas.1702940114 Mock et al. Downloaded by guest on September 30, 2021 To confirm that the Bag6-NLS is a KPNA-binding site and to GST Pulldown MBP Pulldown define the residues involved in KPNA binding, 131 C-terminal residues of Bag6 (Bag6C131) (32) in complex with hexahistidine- MBP•KPNA2 tagged full-length Ubl4A (Bag6C131–6xHis·Ubl4A or Bag6– GST•TRC35 Ubl4A) was purified from E. coli. KPNA2 has been shown to Bag6WT interact with Bag6 (33) and could be stably purified. Mutations 116 1 2 3 4 5 6 7 8 9 1011121314 Lanes previously shown to abrogate Bag6 nuclear localization (31) were 97.4 MBP•KPNA2 introduced at either the first basic cluster (1024SL) or the second 66.2 basic cluster (1045SL) (Fig. S7A). The purified Bag6–Ubl4A variants GST•TRC35 were incubated with either GST·TRC35 or MBP·KPNA2 lacking the 45 importin β binding domain (58–529), and the resulting complexes were isolated using Ni-NTA beads. WT, 1024SL, and 1045SL Bag6- 31 Ubl4A pulled down GST·TRC35 with similar efficiency (Fig. S7B, Ubl4A 21.5 lanes 5–7), demonstrating that these residues are not involved in 14.4 Bag6 binding TRC35. MBP·KPNA2 formed a complex with Bag6 that can be visualized in a pulldown (Fig. S7B, lane 9). Surprisingly, only Fig. 5. TRC35 binding precludes karyopherin α binding to Bag6. Recombi- mutation of the second basic cluster led to disruption of the in- nantly purified Bag6C131–6xHis·Ubl4A (500 pmol) was first incubated with teractionbetweenBag6–Ubl4A and MBP·KPNA2 (Fig. S7B,lane either GST·TRC35 or MBP·KPNA2 (500 pmol). The resulting GST·TRC35–Bag6– 10 vs. lane 11). These results show that the residues required for Ubl4A complex was incubated with glutathione resin beads, and increasing amounts of MBP·KPNA2 (50, 250, or 1,000 pmol) was added. The ability of binding KPNA2 are distinct from those required for binding TRC35. · – · Moreover, unlike what had been previously suggested (31), the Bag6 MBP KPNA2 to displace Bag6 Ubl4A from GST TRC35 was examined by eluting the GST·TRC35 and bound Bag6–Ubl4A from the glutathione resin. NLS is monopartite with the second basic cluster both necessary and The opposite experiment was also carried out using amylose resin beads. sufficient for binding KPNA2. The overlap of the two binding sites provides a mechanism for Bag6 retention in the cytoplasm. If binding of TRC35 to Bag6 prevents KPNA-mediated trans- TRC35 to Ubl4A. The Bag6–TRC35 structure reveals that location, only TRC35-free Bag6 should be able to bind KPNA TRC35 binds this binding partner utilizing conserved patches in and be translocated to the nucleus. One way this could be a distinct fashion (Fig. 1). This led to the expansion of the achieved is if KPNAs have a higher affinity for Bag6 than TRC35 protein–protein interaction network while maintaining BIOCHEMISTRY TRC35, up-regulation of KPNA expression would lead to dis- its crucial interaction with TRC40. – placement of TRC35 from Bag6 and formation of a Bag6 KPNA Our results also show that TRC35 acts as an intermolecular complex. To test this hypothesis, we compared the relative mask to Bag6 NLS in a role similarly performed in other path- binding affinities of TRC35 and KPNA2 to Bag6 using an ex- ways. Examples of other cytosolic retention factor pairs include – • change assay. Bag6 Ubl4A complexes with GST TRC35 were IκB and NF-κB (35) and BRAP2 that retains HMG20A (36). generated and bound to glutathione affinity resin. After washing, Unlike other cytoplasmic retention factors, which have only been increasing amounts of KPNA2 were added. The ability of shown to bind their target NLS-containing proteins for occlusion KPNA2 to displace Bag6–Ubl4A from TRC35 was determined – of the NLS, TRC35 also plays a distinct role in the cytoplasmic by the amount of Bag6 Ubl4A eluted from the resin after in- TA targeting and protein quality control when in complex with cubation. In this case, even at the highest concentration tested Bag6. This dual functionality may be evolutionarily conserved. In (twofold molar excess), there was no significant displacement of yeast, fungal Get4 binds the NTD of Get5, which appears to Bag6 from TRC35 by KPNA2 (Fig. 5, lanes 4–7). When starting contain a functional NLS that directs Get5 to the nucleus during with MBP·KPNA2–Bag6–Ubl4A on amylose beads, adding ex- a Get5-mediated stress response (37). cess TRC35 resulted in the dissociation of the KPNA2–Bag6– These results allow speculation of possible regulatory mech- Ubl4A complex (Fig. 5, lanes 11–14). These results highlight the anism for Bag6 nuclear localization. First, disrupting the Bag6– stability of the TRC35–Bag6 complex and argue against the TRC35 interface by introducing alanine mutations led to ubiq- ability of KPNA regulation as a means for modulating the nu- uitylation of TRC35 (Fig. S5), which is mediated by RNF126. clear pool of Bag6. Knocking down Bag6 has been shown to reduce levels of Discussion TRC35 in HeLa cells (22), suggesting that Bag6 is required for Bag6 is a critical scaffolding factor that forms a stable complex TRC35 stability in vivo. We also showed that KPNA2 has a lower with TRC35 and Ubl4A and is involved in protein targeting, affinity for Bag6 than TRC35 (Fig. 5), which suggests that for protein quality control, and transcription regulation. While there Bag6 to bind KPNA and translocate into the nucleus, it needs to be is ample evidence that Bag6 plays both cytosolic and nuclear free of TRC35. The most likely explanation is that cells regulate roles, it is unclear how the localization of Bag6 is regulated. Here Bag6 localization by modulating Bag6 levels or decreasing TRC35 we report the structure of the Bag6 NLS region bound to its levels. In humans, the Bag6 rs3117582 single nucleotide poly- cytosolic retention factor TRC35, and suggest a mechanism for morphism found in the promoter region of Bag6 likely affects ex- the regulation of Bag6 localization. pression levels and is associated with higher incidence of lung The remarkable conservation of TRC35 both in sequence (Fig. cancer (38–40) and osteoarthritis (41). The localization could also S3) and at the structural level (Fig. S1) from yeast to human can be pretranslationally regulated with differential splicing. In brain be attributed to the role of TRC35 as a hub of protein–protein and breast tissue, for instance, Bag6 isoforms that lack the NLS are interactions in the TRC pathway. The most important interac- expressed at higher levels than isoforms with the NLS (42). tion, based on its conservation, is between TRC35 and TRC40 Our findings establish TRC35 as a universally conserved (Fig. S2). This interaction is critical for the regulation of TA protein in the opisthokont lineage in both structure and function. protein transfer (3, 13) and, as shown in fungal Get4, for release These findings support the model in which TRC35 is an im- of Get3/TRC40 from the ER membrane to restore the cytosolic portant hub of a protein–protein interaction network with its pool of complex ready for TA protein transfer (34). The im- interaction with TRC40 at the core. Higher eukaryotes expanded portance of linking TRC35 homologs to Ubl4A homologs is on this network by the addition of Bag6. Our biochemical studies sustained in humans despite the differences in structural features unequivocally establish TRC35 as a cytosolic retention factor of (15). In metazoans, the component Bag6 acts as a scaffold to link Bag6 and suggest that Bag6 needs to be in excess for nuclear

Mock et al. PNAS Early Edition | 5of6 Downloaded by guest on September 30, 2021 localization. However, how different cell types regulate Bag6 single-wavelength anomalous dispersion. The genomes were surveyed localization, whether the regulation can be mediated by specific using MEME suite (43). The IP assays were carried out using 293T cell ex- stress, and the biological implications of differential distribution tracts. A combination of three siRNAs against RNF126 was added to μ of Bag6 are important questions for future studies. 293T cells. MG132 (10 M) was added to inhibit the proteasome. Cos7 cells were used for localization assays. Materials and Methods ACKNOWLEDGMENTS. We thank Daniel Lin and Jens Kaiser for help with Detailed descriptions of experiments are provided in SI Materials and data processing, Catherine Day for technical support, members of the Methods. Briefly, for crystallization and in vitro binding assays human W.M.C. laboratory for support and useful discussions, Gordon and Betty were subcloned and expressed in Escherichia coli. Proteins were purified Moore for support of the Molecular Observatory at California Institute of using affinity chromatography, anion-exchange, and size-exclusion chro- Technology, and the staff at the Stanford Synchrotron Radiation Light- matography. Truncated fragments of Bag6 and TRC35 were crystal- source for assistance with synchrotron data collection. W.M.C. is supported lized, and the structure was determined using molecular-replacement by NIH Grant R01GM097572.

1. Kalbfleisch T, Cambon A, Wattenberg BW (2007) A bioinformatics approach to 31. Manchen ST, Hubberstey AV (2001) Human Scythe contains a functional nuclear lo- identifying tail-anchored proteins in the . Traffic 8:1687–1694. calization sequence and remains in the nucleus during staurosporine-induced apo- 2. Stefanovic S, Hegde RS (2007) Identification of a targeting factor for posttranslational ptosis. Biochem Biophys Res Commun 287:1075–1082. membrane protein insertion into the ER. Cell 128:1147–1159. 32. Wu YH, Shih SF, Lin JY (2004) Ricin triggers apoptotic morphological changes through 3. Mock JY, et al. (2015) Bag6 complex contains a minimal tail-anchor-targeting module caspase-3 cleavage of BAT3. J Biol Chem 279:19264–19275. and a mock BAG domain. Proc Natl Acad Sci USA 112:106–111. 33. Rouillard AD, et al. (2016) The harmonizome: A collection of processed datasets 4. Schuldiner M, et al. (2008) The GET complex mediates insertion of tail-anchored gathered to serve and mine knowledge about genes and proteins. Database (Oxford) proteins into the ER membrane. Cell 134:634–645. 2016:baw100. 5. Rome ME, Rao M, Clemons WM, Shan SO (2013) Precise timing of ATPase activation 34. Rome ME, Chio US, Rao M, Gristick H, Shan SO (2014) Differential gradients of in- drives targeting of tail-anchored proteins. Proc Natl Acad Sci USA 110:7666–7671. teraction affinities drive efficient targeting and recycling in the GET pathway. Proc 6. Yamamoto Y, Sakisaka T (2012) Molecular machinery for insertion of tail-anchored Natl Acad Sci USA 111:E4929–E4935. membrane proteins into the endoplasmic reticulum membrane in mammalian cells. 35. Beg AA, et al. (1992) I kappa B interacts with the nuclear localization sequences of the Mol Cell 48:387–397. subunits of NF-kappa B: A mechanism for cytoplasmic retention. Genes Dev 6: 7. Vilardi F, Lorenz H, Dobberstein B (2011) WRB is the receptor for TRC40/Asna1-me- 1899–1913. diated insertion of tail-anchored proteins into the ER membrane. J Cell Sci 124: 36. Davies RG, Wagstaff KM, McLaughlin EA, Loveland KL, Jans DA (2013) The BRCA1- – 1301 1307. binding protein BRAP2 can act as a cytoplasmic retention factor for nuclear and 8. Rivera-Monroy J, et al. (2016) Mice lacking WRB reveal differential biogenesis re- nuclear envelope-localizing testicular proteins. Biochim Biophys Acta 1833: quirements of tail-anchored proteins in vivo. Sci Rep 6:39464. 3436–3444. 9. Abell BM, Pool MR, Schlenker O, Sinning I, High S (2004) Signal recognition particle 37. Arhzaouy K, Ramezani-Rad M (2012) Nuclear import of UBL-domain protein Mdy2 is – mediates post-translational targeting in eukaryotes. EMBO J 23:2755 2764. required for heat-induced stress response in Saccharomyces cerevisiae. PLoS One 7: 10. Aviram N, et al. (2016) The SND proteins constitute an alternative targeting route to e52956. – the endoplasmic reticulum. Nature 540:134 138. 38. Etokebe GE, et al. (2015) Association of the FAM46A gene VNTRs and BAG6 rs3117582 11. Abell BM, Rabu C, Leznicki P, Young JC, High S (2007) Post-translational integration of SNP with non small cell lung cancer (NSCLC) in Croatian and Norwegian populations. tail-anchored proteins is facilitated by defined molecular chaperones. J Cell Sci 120: PLoS One 10:e0122651. 1743–1751. 39. Zhao J, Wang H, Hu W, Jin Y (2014) Effect of HLA-B-associated transcript 3 polymor- 12. Itakura E, et al. (2016) Ubiquilins chaperone and triage mitochondrial membrane phisms on lung cancer risk: A meta-analysis. Med Sci Monit 20:2461–2465. proteins for degradation. Mol Cell 63:21–33. 40. Chen J, Zang YS, Xiu Q (2014) BAT3 rs1052486 and rs3117582 polymorphisms are 13. Gristick HB, et al. (2014) Crystal structure of ATP-bound Get3-Get4-Get5 complex associated with lung cancer risk: A meta-analysis. Tumour Biol 35:9855–9858. reveals regulation of Get3 by Get4. Nat Struct Mol Biol 21:437–442. 41. Etokebe GE, et al. (2015) Susceptibility to large-joint osteoarthritis (hip and knee) is 14. Bozkurt G, et al. (2010) The structure of Get4 reveals an alpha-solenoid fold adapted associated with BAG6 rs3117582 SNP and the VNTR polymorphism in the second for multiple interactions in tail-anchored protein biogenesis. FEBS Lett 584: of the FAM46A gene on 6. J Orthop Res 33:56–62. 1509–1514. 42. Luce MJ, Akpawu AA, Tucunduva DC, Mason S, Scott MS (2016) Extent of pre-trans- 15. Chartron JW, Suloway CJ, Zaslaver M, Clemons WM, Jr (2010) Structural character- lational regulation for the control of nucleocytoplasmic protein localization. BMC ization of the Get4/Get5 complex and its interaction with Get3. Proc Natl Acad Sci Genomics 17:472. USA 107:12127–12132. 43. Bailey TL, et al. (2009) MEME SUITE: Tools for motif discovery and searching. Nucleic 16. Chang YW, et al. (2010) Crystal structure of Get4-Get5 complex and its interactions Acids Res 37:W202–W208. with Sgt2, Get3, and Ydj1. J Biol Chem 285:9962–9970. 44. Kabsch W (2010) Xds. Acta Crystallogr D Biol Crystallogr 66:125–132. 17. Mariappan M, et al. (2010) A ribosome-associating factor chaperones tail-anchored 45. Collaborative Computational Project, Number 4 (1994) The CCP4 suite: programs for membrane proteins. Nature 466:1120–1124. protein crystallography. Acta Crystallogr D Biol Crystallogr 50:760–763. 18. Wang Q, et al. (2011) A ubiquitin ligase-associated chaperone holdase maintains 46. Winn MD, et al. (2011) Overview of the CCP4 suite and current developments. Acta polypeptides in soluble states for proteasome degradation. Mol Cell 42:758–770. – 19. Sebti S, et al. (2014) BAT3 modulates p300-dependent acetylation of p53 and auto- Crystallogr D Biol Crystallogr 67:235 242. phagy-related protein 7 (ATG7) during autophagy. Proc Natl Acad Sci USA 111: 47. Bricogne G, Vonrhein C, Flensburg C, Schiltz M, Paciorek W (2003) Generation, rep- 4115–4120. resentation and flow of phase information in structure determination: Recent de- 20. Sasaki T, et al. (2007) HLA-B-associated transcript 3 (Bat3)/Scythe is essential for p300- velopments in and around SHARP 2.0. Acta Crystallogr D Biol Crystallogr 59: – mediated acetylation of p53. Genes Dev 21:848–861. 2023 2030. 21. Wakeman TP, Wang Q, Feng J, Wang XF (2012) Bat3 facilitates H3K79 dimethylation 48. McCoy AJ (2007) Solving structures of protein complexes by molecular replacement – by DOT1L and promotes DNA damage-induced 53BP1 foci at G1/G2 cell-cycle phases. with Phaser. Acta Crystallogr D Biol Crystallogr 63:32 41. EMBO J 31:2169–2181. 49. Adams PD, et al. (2010) PHENIX: A comprehensive Python-based system for macro- – 22. Krenciute G, et al. (2013) Nuclear BAG6-UBL4A-GET4 complex mediates DNA damage molecular structure solution. Acta Crystallogr D Biol Crystallogr 66:213 221. signaling and cell death. J Biol Chem 288:20547–20557. 50. Emsley P, Lohkamp B, Scott WG, Cowtan K (2010) Features and development of Coot. 23. Wainright PO, Hinkle G, Sogin ML, Stickel SK (1993) Monophyletic origins of the Acta Crystallogr D Biol Crystallogr 66:486–501. metazoa: An evolutionary link with fungi. Science 260:340–342. 51. Eme L, Sharpe SC, Brown MW, Roger AJ (2014) On the age of eukaryotes: Evaluating 24. Baldauf SL, Palmer JD (1993) Animals and fungi are each other’s closest relatives: evidence from fossils and molecular clocks. Cold Spring Harb Perspect Biol 6:a016139. Congruent evidence from multiple proteins. Proc Natl Acad Sci USA 90:11558–11562. 52. Bailey TL, Gribskov M (1998) Combining evidence using p-values: Application to se- 25. Xu Y, Anderson DE, Ye Y (2016) The HECT domain ubiquitin ligase HUWE1 targets quence homology searches. Bioinformatics 14:48–54. unassembled soluble proteins for degradation. Cell Discov 2:16040. 53. Altschul SF, Gish W, Miller W, Myers EW, Lipman DJ (1990) Basic local alignment 26. Hessa T, et al. (2011) Protein targeting and degradation are coupled for elimination search tool. J Mol Biol 215:403–410. of mislocalized proteins. Nature 475:394–397. 54. Gietz RD, Schiestl RH (2007) Large-scale high-efficiency yeast transformation using 27. Rodrigo-Brenni MC, Gutierrez E, Hegde RS (2014) Cytosolic quality control of mis- the LiAc/SS carrier DNA/PEG method. Nat Protoc 2:38–41. localized proteins requires RNF126 recruitment to Bag6. Mol Cell 55:227–237. 55. Schrödinger, LLC (2015) The PyMOL Molecular Graphics System (Schrödinger, LLC, 28. Melikova MS, Kondratov KA, Kornilova ES (2006) Two different stages of epidermal New York), Version 2.0. growth factor (EGF) receptor endocytosis are sensitive to free ubiquitin depletion 56. Pettersen EF, et al. (2004) UCSF Chimera–A visualization system for exploratory re- produced by proteasome inhibitor MG132. Cell Biol Int 30:31–43. search and analysis. J Comput Chem 25:1605–1612. 29. Hoelz A, Debler EW, Blobel G (2011) The structure of the nuclear pore complex. Annu 57. Katoh K, Standley DM (2013) MAFFT multiple sequence alignment software version 7: Rev Biochem 80:613–643. Improvements in performance and usability. Mol Biol Evol 30:772–780. 30. Pumroy RA, Cingolani G (2015) Diversification of importin-α isoforms in cellular 58. Bailey TL, Elkan C (1994) Fitting a mixture model by expectation maximization to trafficking and disease states. Biochem J 466:13–28. discover motifs in biopolymers. Proc Int Conf Intell Syst Mol Biol 2:28–36.

6of6 | www.pnas.org/cgi/doi/10.1073/pnas.1702940114 Mock et al. Downloaded by guest on September 30, 2021